JPH07262074A - Cache management method, computer file system and cache device - Google Patents

Abstract

PURPOSE: To make the cache of the storing means of a client operable when the object is disconnected from a server by confirming whether or not data related to a file system object exist by testing the cache and responding to a file system object request when the data exist. CONSTITUTION: A request 320 is vicariously received by means of a VFS interface 328, delivered to a distributed file system client 322 residing in the kernel address space 312 of the client 322, and further delivered to a distributed file system server 324 at a server kernel 316 in the form of a message 326 through a network. The server 324 accesses a storage device 390 by using the VFS interface 328. The client 322 inspects a cache for the presence of effective data in the cache before requesting data from the server 324.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an information processing system for managing data. More specifically, the present invention relates to client data management in distributed file systems. More specifically, the present invention relates to caching and logging in distributed file system clients that can support client operations on connect or disconnect.

[0002]

BACKGROUND OF THE INVENTION Computer workstations have increased in capacity and storage capacity. In essence, a workstation has 1 unit for performing one or more separate tasks.
Used by human operators. Workstations have been deployed to many users within an organization and the number has increased, which has created the need to communicate between workstations and share data between users. As a result, a distributed file system architecture as shown in FIG. 1 has been developed.

In FIG. 1, a plurality of workstations are interconnected by a local area network (LAN). A wide area network (WAN) may also be used to interconnect multiple local LANs. LAN and WAN form a single logical network. In FIG. 1, the workstation is the server S1,
Shown as S2, S3 and clients C1, C2, C3. Whether a workstation is called a client or a server depends on the function it performs in the network. In some cases, a particular workstation doubles as both a client and a server. To implement a distributed file system, there must be at least one server and at least one client.

Traditionally, client workstations have always been connected to a LAN to receive LAN services and share resources such as files. In other words, when disconnecting from the LAN, the workstation
It meant being cut off from resources. for that reason,
The disconnected client could not continue without the necessary files and programs copied to the client side before disconnecting.

The increased capabilities of portable workstations have made the computing environment or "mobile" computing in an isolated state feasible.
You can now buy a portable workstation with the same processor, memory, and disk storage capacity as a desktop or deskside unit. However, current portable workstations or computers are subject to the same restrictions as above when disconnected. In other words, if you do not make a copy of the file or program before disconnecting, you cannot use any shared resources. When reconnecting the workstation, the user must manually reconcile the modified file on the portable device with the file on the network. In FIG. 1, a client C1 102 has a disconnectable portable workstation connected to the network by a dotted line.
As shown.

The sharing of data files over networks such as that shown in FIG. 1 has evolved over time.
In its simplest form of sharing, a client can request data from a file on the server side. The required data is sent to the client processor and any changes or modifications made to that data are returned to the server. Appropriate locks are set to prevent the second client from changing the data in the file held by the first client.

The distributed file system enhances the file sharing function by adding a mechanism for more efficiently distributing data to a plurality of clients and more efficiently controlling file sharing. Distributed file
There are many systems. A typical distributed file system is And sold by Transarc.
There is a rew file system (AFS).

AFS enhances the efficiency of distributed file processing by creating a file cache on the client side that accesses server data. This cache is referenced by the client application and data is retrieved from the server only if a cache miss occurs. Caching data reduces network traffic and client response time.

AFS cache coherency is based on the server's callback system. The server is
Notify each client whose particular data has been cached of the condition that invalidates the client's cached data. Invalidated data is removed from the cache and must be reacquired from the server if necessary.

Disconnecting a workstation connected to the AFS results in loss of distributed file access by the client. If the cache is invalidated by a disconnect, the client automatically clears all callbacks.

Distributed File System (DFS) is an A
Second to FS, it supports the OSF Distributed Computing Environment (DCE). DFS uses a server-based token mechanism to ensure client cache coherency. The client gets a "read" token so that the data in its cache is valid. When a client modifies the data in a file, it gets a "write" token for that data. Granting a "write" token to a client such as C1 invalidates a "read" token for the same data that exists on all other clients. Invalidating the token invalidates the cached data for these clients. Upon disconnecting from the DFS system, the client loses outstanding tokens. The client will not be able to read or write to cached data due to the missing token.

Carnegie Mellon University has endeavored to support detachable workstations connected to the AFS network. This CODA project aims to provide a distributed file system with constant data availability. It achieves this goal by copying the data to multiple servers and supporting disconnect operations by clients. "Coda: A highly Avai
lable File System fora Distributed Workstation Env
ironment "(IEEE Transactions, Vol. 39, No.4, 19
See April 1990).

The distributed operation of CODA is achieved by optimistic data replication control. With optimistic control, many clients can read and write data even when disconnected. Inconsistencies in the data remain to be identified and resolved later. With pessimistic replication control,
Avoid all conflicts by limiting reading and writing to a single partition. Disconnecting prevents a client from determining if another client can access the data. Also, under pessimistic replication control, the client may not be able to write. "Disconnected Oper
ation in the Coda File System "(J. Kistler, M Saty
anarayanan, Carnegie Mellon University, Proceeding
s of the 13th ACM Symposium on Operating Systems P
rinciples, October 1991).

The distributed operation of CODA allows a client to continue to access and update data held in its client-side cache. Only in this case will a file operation fail because a cache miss will prevent the client from accessing the data from the server. AFS and CODA
Caches the entire file. CODA updates the file during the disconnect operation and keeps a log of all changes made to that data. When reconnected, the logged transaction will
It is the client's responsibility to update all server-side copies of that data by applying it to the file. Data inconsistencies are dealt with by deferring resolution for manual intervention.

The main drawback of CODA is that it only supports connections to CODA servers. Like AFS, CODA is a distributed file system protocol. CODA relies on the fact that the server (s) and all clients use the same distributed file system protocol. Therefore, all clients and servers in the network must be changed to the CODA protocol to support disconnect operations.

Another approach is "Disconnected Op"
eration for AFS "(LB Huston and P. Honeyman, C
enter for Information Technology Integration, The
University of Michigan, published in the Proceedin
gs of the USENIX Mobile and Location-Independent C
Hust at the omputing Symposium, August 1993).
Proposed by on and Honeyman. The client system proposed by Huston and Honeyman can connect to a standard AFS server without modification to the server. Client-side code modifications support disconnect operations and reconnects. Reconnection and reconciliation is achieved by logging all transactions during disconnects. Every read, write, or update is recorded in the transaction log and this transaction log is replayed to the server on reconnect.
This approach has the advantage of supporting standard AFS servers. However, it relies on the existence of the AFS distributed file system protocol and therefore does not work on servers other than the AFS server. This fully logged approach also compromises system efficiency. Also, log size can lead to client-side disk space issues and throttling delays on systems of all sizes.

Therefore, there is a technical problem in providing a distributed file system in which a disconnected client can continuously operate by performing efficient adjustment when reconnecting. In addition, many distributed file systems
It is also a technical problem to create a distributed file system client that cannot be used in the protocol. Finally, there is also the technical problem of creating a distributed file system client independent of operating system syntax for object or pathname resolution.

[0018]

SUMMARY OF THE INVENTION Therefore, one object of the present invention is to provide a distributed file system client which is operable when disconnected from a server.

Another object of the present invention is to provide a disconnectable client that efficiently logs transactions for coordination during reconnection.

Another object of the present invention is to provide a file system client equipped with an efficient cache function in order to reduce the memory capacity and disk space required for cache and to improve the speed and efficiency of cache processing. To provide.

Another object of the present invention is to provide a distributed file system client operable with a plurality of distributed file system protocols.

Another object of the present invention is to
To provide a distributed file system client that is system independent.

Another object of the present invention is to improve cache efficiency by aggressively caching decomposed object names.

Another object of the present invention is to provide a cache management system that allows each file system object to have a plurality of parent objects that are independent of the object name and each have different object name syntax rules. To do.

The above and other objects, features and characteristics of the present invention,
And advantages will be apparent from the detailed description of the embodiments of the present invention shown in the accompanying drawings. However, in the accompanying drawings, the same reference numerals represent the same parts of the present invention.

[0026]

SUMMARY OF THE INVENTION The present invention seeks to provide a distributed file system that supports both connect and disconnect client operations. This client file system can connect to multiple file system architectures and supports efficient transaction logging.

The present invention is directed to a method of managing a file system cache of a client computer system operating under a first operating system. The method comprises intercepting an operating system request for a file system object of a distributed file system, transforming the request to remove operating system dependent syntax, and storing means of the client. Test the cache of the file system
It includes the steps of checking if data exists for the object, and responding to the file system object request if cached data exists.

[0028]

DETAILED DESCRIPTION FIG. 2 illustrates a typical client or server workstation configuration. The present invention is
It is preferably implemented on a workstation such as a BM RISC System / 6000 workstation or an IBM PS / 2 personal computer. The client and server workstations are shown in FIG.
Token Ring or Ethernet LA as shown in
Interconnected by N. Workstations connected to the LAN can include portable workstations such as the C1, disconnect from the LAN,
It can also be used independently of the LAN. It should be noted that many other configurations of workstation hardware and LAN types can be used within the scope of the present invention.

A typical client or server workstation 200 has a processor 210, system memory 214, and non-volatile storage 212 such as a hard disk drive, diskette drive, or optical disk storage. The processor accepts input from I / O devices such as keyboard 224 and pointing device 226 via I / O controller 218. This system includes a graphic controller 220.
The graphic information is presented to the operator on the display device 222 driven by. This workstation is
It is connected to a network (not shown) via the network interface adapter 216.

The preferred embodiment of the present invention is implemented by a computer process running in the memory and processor of workstation 200. A computer program product embodying the present invention may be stored in non-volatile storage 212, such as storage on a tape, diskette, or compact disc read only memory (CD-ROM) device.

The operation of the prior art distributed file system is illustrated with respect to FIG. FIG. 3 shows a client workstation 302 and a server workstation 304 interconnected by a network interface 306. A request by user process 308 for a particular file or for information about a file that resides on disk storage 390 is processed as follows. However, both client 302 and server 304 have user address spaces 310, 314 and kernel address space 31.
Note that it is divided into 2,316. Interface between application and kernel (application program interface, ie A
Called PI) is "Vnodes:" written by SR Kleiman.
An Architecture for Multiple File System Types in
Sun UNIX "(Summer Usenix Conference Proceedings,
VFS / vnode A defined in 1986)
It is PI. Of course, other interfaces could be used. In this VFS / vnode interface, multiple virtual file systems can exist. This makes the distributed file system one of the virtual file systems for client workstations.

The user process 308 uses the storage device 39.
Request data file on 0. Request 320 is V
It is intercepted by the FS interface and passed to the distributed file system client 322 which resides in the client's kernel address space 312. The distributed file system client manages communication with the server on the network. The request is passed in the form of a message 326 over the network 306 to the distributed file system server 324 at the server kernel 316. Distributed File System Server 324 is VFS
The interface 328 is used to access the storage device 390 by accessing the server-side local physical file system including the storage device 390. The requested data has been acquired and distributed file system
It is passed from the server 324 to the distributed file system client 322, which in turn requests the requesting user process 30.
Pass the data to 8.

Distributed File System Client 3
22 may include a data cache (not shown) to cache the data requested by the server. The distributed file system client checks the cache to see if it has valid data before requesting it from the server.

A preferred embodiment of the present invention is shown in FIG. The client workstation has a user address space 402 and a kernel address space 40.
Have 4. The improved cache management program 406 of the present invention preferably operates in the user address space 402, but may be implemented in the kernel address space 404.

User programs 408, 410, 41
2 is a file system request 414, 416, 418
Give out. User program n is shown as using a logical file system 402, which issues a file system request 420. File system requirements are VFS / v
It is compatible with the nodes interface, but other interfaces could be used within the scope of the invention. The present invention can be used with any logical file system that is compatible with the VFS / vnode API or, in alternative embodiments, with other APIs.

The VFS interface sends file system requests to the cache manager 406. The cache management program 406 stores local information about the distributed file system accessed by the client. When possible, the cache manager services file system requests without accessing the server. If the requested information is in the cache and is still valid, the cache manager can serve the file system.
The cache management program 406 will be described in detail below.

The cache manager 406 responds to file system requests by either resident data in the cache manager memory or by accessing the local physical file system 420,
Or use distributed file system access 422. The cache manager stores a local copy of the distributed file in the local physical file system for quick access. Distributed file system access 422 issues the necessary commands to distributed file system server 324, as described above. Although the preferred embodiment caches the entire file, the invention is not limited to a full file cache.

The cache manager 406 and distributed file system access 422 are independent of the operating system and distributed file system access protocol. Independence from the operating system is achieved by supporting predefined interfaces. Supported interfaces include VF for file system operations
S + / Vnodes interface, IBM SOMo
persistent interfaces, such as those provided by the objects persistence framework, a local file system interface (LS) for cache files.
I), various distributed file system protocols, cache synchronization interfaces, log replay protocols, etc. The cache manager is independent of any operating system syntax. Filenames are converted to be independent of filename component separators and are considered to contain no reserved characters or words. Pathname resolution is also independent of operating system syntax. Thus, a particular file system object can be referenced by two operating systems whose syntax rules conflict. For example, the cache management program 406 uses I
BM OS / 2 operating system or IB
Works on M AIX operating system,
It can be implemented to work on other operating systems with only minor modifications.

The cache management program of the present invention is independent of the distributed file system protocol.
Distributed File System Access 422 uses a distributed computing environment (DCE) server with the Open Software Foundation (OSF) or IBM
The LAN Server program product can be used to access remote servers. Other remote file access protocols can be used with minor modifications. Protocol dependent tasks are handled by the synchronization mechanism and the log replay mechanism.

The cache management program 406 manages the client cache 409. Cache 4
In connection with 09, the cache management program stores volume information 410 and modification log 412. Volume information 410 includes information about the file system accessible to the client workstation via the cache management program. In the preferred embodiment, a remote file system is attached to the client by volume. Each volume has a volume item in the volume information database. Thus, a remote file system can be used by a client as a UNIX type file system (eg, mounted as a particular file system / rfs) or as a DCE drive letter (eg, drive G: \). Connected to workstation. All information about the remote file system and its status can be found in the volume information database 410.
Be stored in. It is also possible to connect multiple remote file systems simultaneously to a single client.

The structure of the cache 409 is shown in more detail in FIG. Cache 409 stores a single cache for file path names, file system object data, and file system object status information. The cache is stored in the persistent heap. That is, the cache is regularly written in the nonvolatile storage device, and the cache can be restored from the nonvolatile storage device in the event of a power or system failure. Consolidating the caches together reduces the overall required cache memory capacity, which improves cache performance. Prior art systems required three hash tables, whereas a single cache requires only one. All caching can be accomplished with a set of code that stores various types of file system objects, rather than the traditional mechanism of a separate code for each type of cache. cache·
The file system object is shown generally at 502. This file system object 5
02 is used for both the connection and disconnection operations of the cache management program.

File system object 502
Is a data structure that contains the information needed to access a remote file or a local cached copy of a remote file. A file system object entry is created in the cache for each remote file or directory referenced by the client workstation. The file system object contains a VFS + / vnode identifier 504 and a file identifier 506. The file identifier 506 is
Includes volume, server, and file data.
The VFS + / vnode identifier is VFS / Vnode.
It contains information defined by the s standard specification.
Because it contains such information, all VFS / V
It can fully support the node interface. In the present invention, additional information is added to this data structure in order to support the cache enhancement and the operation at the time of disconnection. Volume item 508 is a pointer to volume item 545 in volume information database 410 for the volume to which the file belongs. State variables 510 and status structure 512 are those that are stored to provide cache status information about the data.
In the preferred embodiment, implemented in an object oriented environment using tools such as the C ++ programming language, this situation structure is defined as an object class. The status structure for a particular operating system file (eg OS / 2 or AIX) is defined by creating a subclass of the status structure class. This subclass overrides the status structure class with the exact content needed to define the file status.

The file system object stores cache coherency information 514. This information is used to ensure consistency between the cache and the server and server-side replicas. Access control list (AC
List 516 in L) defines permissions for file system access and modification according to the IEEE POSIX standard. Other flag information is stored in 518. The mount state of the volume is stored in 520 like the parent pointer 522 and the priority information 524. Each file system object is assigned a priority that is used in determining which object to remove to make room for a new object.

File system data is accessed via the file / directory / symbol link pointer 526. This pointer allows access to three types of cached data, depending on the file system object type. data·
The file system object for the file is given a pointer to the cache file 540,
This cache file 540 is a local cache data file in the local file system 420.
Refers to a copy of a file. The directory pointer points to the cache directory 542. This cache directory 542 is built on the aggressive cache scheme. If directory information is requested for a particular directory, that directory and all of its immediate child directories are returned. This increases the probability that information about one or more child directories will be requested immediately after the parent directory information, thus reducing the number of cache misses. The directory cache provides the information needed for pathname resolution. That is, a particular file system path request is mapped to a particular data file. Finally, pointer 526 may point to cache symbol link information 544. This cached information enables the elucidation of symbolic links.

At 528, items for various counters, locks, and handles are provided, and at 532, an indication of clean status is provided. Clean status indicator 53
2 is important for the operation at the time of disconnection and is used to indicate the adjustment status.

It is important to note that the names of file system objects are not included in data structure 502. This allows file system objects to be referenced from different sets of object names without conflicts. Each file system object can have multiple parent objects. Each parent is an object management group (OM
Different naming conventions may be met, such as G) naming specifications, or AIX, OS / 2, and other operating system naming specifications. As a result, the specific file object employee_list is
From 2, set c: \ employee_list to A
It can be accessed from IX as / usr / employee_list.

The cache management program 406 supports both connection and disconnection operations. During connect-time operation, cache manager 406 responds to file system requests by the cache, or by remote file access if necessary. Upon disconnection, the cache manager manages the fix log 546 for all file system fixes in response to requests from the cache.

FIG. 6 shows one embodiment of the cache management program according to the present invention. This example is based on IBM
It is designed for use with the OS / 2 operating system. File system requirements are "
issued by the system as a "doscall" command 602. I for all file system requests
The FS file system 604 corresponds. IFS
It provides the functions of a logical file system. Mount various types of file systems,
IFS can support these file systems. As the mounted file system,
FAT file system 60 common to DOS systems
6 and the OS / 2 HPFS file system 608 and the like. In a LAN environment, the LAN requester 6
10 exists as another file system, where file access is passed to the LAN server for treatment. The cache management program according to the present invention can be mounted as another type of file system 612. The cache management program 612 is passed all remote file system access. The cache manager tests the cache 614 (in the user address space) to determine if the requested information is already cached. If cached, that information is provided to the application, including provisioning and modification of cached data held in the local file system 606 or 608, if necessary. If the file system data is not found, the request is passed to the LAN requester 610 and the LAN requester accesses the required data on the LAN server. When the remote data arrives, it is placed in the cache for future reference.

A flow chart showing the flow of processing of the file write command is shown in FIG. Other file system commands are implemented as well. The process starts at 702 and is 7
A file write command is issued at 04. The cache manager performs a test at 706 to determine if the requested file is in the cache. If the file exists, then at 708, the cached copy of the file is updated. The cache is then tested at 710 to determine if the volume is currently attached. The connection status is stored in the volume information database 410. If the volume is connected, the changes are propagated to the server at 712,
The process ends at 722. If the volume is not connected, the contents of the change are recorded in the correction log at 714.
The process ends at 22.

If the requested file does not exist in the cache, then the cache manager next returns 716.
To determine if the requested volume is attached. 718 if connected
The file is requested from the remote file system and placed in the cache at step 708 and processing resumes at step 708. If the volume is not connected, a fault signal is returned to the application at 720.

The process of logging the modification details in the modification log 546 (step 714, etc.) is accomplished by logging the object. The problem domain is represented by the client's log modification class and a separate class for each type of transaction logged. Since different objects are defined for each operation type, it is likely that there will be one each for writing files, creating files, and removing files, for example. When the cache manager logs the changes in the disconnect on operation mode at 714, a log object type corresponding to the requested operation is created (eg, file write). The logging method is called to log this object. As such an object method, there is a method of compressing and optimizing a log for reproduction. Logging certain objects may also remove previously logged objects. For example, if you create a file and later remove it, the file removal object removes the file creation log entry and the file modification log entry and ends the file removal object without logging the file removal entry. This system allows the cache to synchronize with the remote file system quickly and efficiently when reconnecting. The information in the logging object eliminates the need to process or compress the log prior to replay.

From the above description, it will be understood that various modifications and changes can be made in the preferred embodiment of the present invention without departing from the true spirit of the invention. The above description is for purposes of illustration only and should not be construed in a limiting sense. The scope of the invention should be limited only by the terms set forth in the claims.

In summary, the following matters will be disclosed regarding the configuration of the present invention.

(1) A method of managing a file system cache of a client computer system operating under a first operating system, said method determining a file system object in a distributed file system. operating·
Intercepting system requests, transforming the request to remove operating system dependent syntax, testing a cache of the client's storage, and writing a file based on the transformed request A method comprising: checking if system object data is present, and responding to the file system object request if cache data is present. (2) performing a test to determine if a connection to the distributed file system exists, and if no cached data exists and a connection exists,
Generating a request from the distributed file system for the file system object regardless of the distributed file system protocol; sending the request to the distributed file system; and having cached data And rejecting the file system object request when there is no connection.
The method according to (1) above. (3) The step of sending the request includes a distributed file for the requested file system object.
Determining a system protocol, converting the operating system independent request into a request conforming to the determined distributed file system protocol, and converting the converted request to the distributed file system The method according to (2) above, further comprising the step of: (4) If the test step determines that there is no connection to the distributed file system, then always perform the subsequent steps and test the file system object to determine the request type. Determining, activating a request logging object corresponding to the request type, invoking a log modification method to log the file system object request, and a log corresponding to the request type By executing the optimization process,
Further comprising logging the file system object request. (5) A file system object
Cache means for storing data, and cache management means for using said cache means to satisfy operating system requests for file system objects
Cache management means independent of system syntax and distributed file system protocol; network means for connecting the cache management means to a file system object server in a detachable manner; A computer file including remote file access means for accessing file system object requests via
system. (6) The cache means is the file system
(5) further comprising object modification log recording means for collecting change information for each operating system request processed by the cache management means when disconnected from the object server. ) Described in the system. (7) In a method of caching a name reference to a distributed file system by a client system, the method converting the name reference to be independent of separator syntax, and from the distributed file system to the syntax. Requesting independent name resolution,
Resolving the syntax-independent name, and caching the resolved name and all names of child components immediately below each path component of the resolved name on the client system. How to include. (8) Storage means for storing a plurality of file system object data items (entries), and first linking means for linking each of the file system object data items to a file volume And linking each of said file system data objects either anywhere in local data storage, to a directory cache in local memory, or to a symbolic link in local memory. A computer system cache device for storing file system information, including a second linking means for storing the file system information. (9) The file system object data item is independent of any file system object name and can be referenced by a plurality of parent file system objects each having a different naming syntax. The computer system cache device according to (8) above. (10) All changes made to the objects referenced by the plurality of file system object data items when the cache is disconnected from the file volume
The computer system cache device according to (8) above, further comprising a log recording unit for recording a log for each volume. (11) In the above (8), further comprising interface means for accepting an operating system request for file system object data and converting the request into an operating system independent request. Computer system cache device as described. (12) All changes made to the objects referenced by the plurality of file system object data items while the cache is disconnected from the file volume are stored in each file.
The computer system cache device according to (9) above, further comprising a log recording unit for recording a log for each volume. (13) In (9) above, further including interface means for accepting an operating system request for file system object data and converting the request into an operating system independent request. Computer system cache device as described. (14) Means for the directory cache means to store resolved directory entries for requested file system data objects, and each path component of the requested file system data objects. Means for storing resolved directory entries for objects in the hierarchy immediately below the computer system cache device according to (11) above. (15) means for the directory cache means to store resolved directory entries for requested file system data objects, and each path component of the requested file system data objects Means for storing resolved directory entries for objects in the hierarchy immediately below the computer system cache device according to (13) above.

[Brief description of drawings]

FIG. 1 is a diagram illustrating a distributed network environment in which the present invention is implemented.

FIG. 2 is a block diagram of a workstation incorporating one embodiment of the present invention.

FIG. 3 is a block diagram of a prior art distributed file system.

FIG. 4 is a block diagram showing functional components of an embodiment of the present invention.

FIG. 5 is a diagram showing a data structure of a preferred embodiment of the present invention.

FIG. 6 is a block diagram of an alternative embodiment of the present invention.

FIG. 7 is a flow chart showing the processing steps of one embodiment of the present invention.

[Explanation of symbols]

 100 LAN / WAN 102 Client C1 104 Server S1 106 Client C2 108 Client C3 110 Server S3 112 Server S2

 ─────────────────────────────────────────────────── ─── Continued Front Page (72) Inventor Sevenem Jersey United States 78758 Austin Cherley Drive, Texas 12004 (72) Inventor Amal Ahmed Sherheen United States 78759 Austin Pathfinder Drive, Texas 6321 (72) Invention Richard Byron Ward United States 78726 Austin, Texas Appletree Lane 11208

Claims (15)

[Claims] 1. A method for managing a file system cache of a client computer system operating under a first operating system, the method comprising: operating on a file system object in a distributed file system. Intercepting system requests, transforming the requests to remove operating system dependent syntax, testing a cache of the client's storage, and based on the transformed requests File system
Checking if object data exists, and if cached data exists, the file
Responding to a system object request. 2. A step of performing a test to determine if a connection to the distributed file system exists, and a distributed file system protocol when cache data is not present and the connection is present. Regardless of the above, generating a request for the file system object from the distributed file system, sending the request to the distributed file system, no cached data and no connection The method of claim 1, further comprising: rejecting the file system object request if not. 3. The step of sending the request determines the distributed file system protocol for the requested file system object; and the request independent of the operating system.
3. The method according to claim 2, further comprising the steps of: converting the request that conforms to the determined distributed file system protocol, and transmitting the converted request to the distributed file system. Method. 4. If the test step determines that there is no connection to the distributed file system, then the following step must be performed, and the file system object must be tested.
Determining a request type; activating a request logging object corresponding to the request type; invoking a log modification method to log the file system object request; Logging said file system object request by performing a corresponding log optimization process. 5. A cache means for storing file system object data in a recording means, and a file system using the cache means.
A cache management means for responding to an operating system request for an object, the cache management means being independent of operating system syntax and distributed file system protocol, and the cache management in a detachable manner. A computer file system including network means for connecting the means to a file system object server and remote file access means for accessing file system object requests via said network means. 6. An object modification log for collecting change information for each operating system request processed by the cache management means when the cache means is disconnected from the file system object server. System according to claim 5, characterized in that it further comprises recording means. 7. A method of caching a name reference to a distributed file system by a client system, the method converting the name reference to be independent of separator character syntax; and from the distributed file system. Requesting syntax-independent name resolution, resolving the name independent of syntax, resolving the name, and all names of child components immediately below each path component of the resolving name. On the client system. 8. A storage means for storing a plurality of file system object data items (entries), and a first means for linking each of said file system object data items to a file volume. Linking means and each of said file system data objects anywhere in local data storage,
A computer system cache device for storing file system information comprising a directory cache in local memory or a second linking means for linking to either a symbolic link in local memory. 9. The file system object
9. The computer of claim 8, wherein the data item is independent of any file system object name and can be referenced by multiple parent file system objects, each having a different naming syntax. -System cache device. 10. The plurality of files when the cache is disconnected from the file volume.
9. The method of claim 8 further comprising logging means for logging, for each file volume, all changes made to the object referenced by the system object data item. Computer system cache device. 11. The method of claim 8 further comprising interface means for accepting operating system requests for file system object data and converting the requests into operating system independent requests. Computer described in
System cache device. 12. The plurality of files when the cache is disconnected from the file volume.
10. The method of claim 9, further comprising logging means for logging, for each file volume, all changes made to the object referenced by the system object data item. Computer system cache device. 13. An interface means for accepting an operating system request for file system object data and converting the request into an operating system independent request. Computer described in
System cache device. 14. The directory cache means for storing resolved directory entries for a requested file system data object, and each path of the requested file system data object. 12. A computer system cache device according to claim 11, including means for storing resolved directory entries for objects in the hierarchy immediately below the component. 15. A means for the directory cache means to store resolved directory entries for a requested file system data object, and each path of the requested file system data object. 14. A computer system cache device according to claim 13 including means for storing resolved directory entries for objects in the hierarchy immediately below the component.